Create a notebook that examines the performance of your formulation.
Your notebook will probably have a text section near the top that answers all of the questions below. After that, it will show the commands you ran to gather the data that caused you to answer in that way.
Across at least two (and ideally three) different problem instances, show us the following:
What's the balance of grounding versus solving time?
How many Variables and Constraints (from "--stats") are there?
In grounding, how does the time scale with the number of domain objects in the instance? (Some polynomial in terms of high level parameters like "number of characters" in Ensemble; don't worry about fitting constants)
Which are the rules in your formulation that are responsible for most the grounding effort and how does your polynomial from the previous item arise from those rules?
In solving, how much time is spent finding a first model, a best model, and finally proving that optimal? (Not everyone is using optimization.) Did you do anything interesting to speed up solving? (Adjusting heuristics, changing optimization strategy, using parallelization.)
Can't change a rubric once you've started using it.